Novel function of vitamin E in regulation of zebrafish (Danio rerio) brain lysophospholipids discovered using lipidomics.

We hypothesized that brains from vitamin E-deficient (E-) zebrafish (Danio rerio) would undergo increased lipid peroxidation because they contain highly polyunsaturated fatty acids, thus susceptible lipids could be identified. Brains from zebrafish fed for 9 months defined diets without (E-) or with (E+) added vitamin E (500 mg RRR-α-tocopheryl acetate per kilogram diet) were studied. Using an untargeted approach, 1-hexadecanoyl-2-docosahexaenoyl-sn-glycero-3-phosphocholine [DHA-PC 38:6, PC 16:0/22:6]was the lipid that showed the most significant and greatest fold-differences between groups. DHA-PC concentrations were approximately 1/3 lower in E- (4.3 ± 0.6 mg/g) compared with E+ brains (6.5 ± 0.9 mg/g, mean ± SEM, n = 10 per group, P = 0.04). Using lipidomics, 155 lipids in brain extracts were identified. Only four phospholipids (PLs) were different (P < 0.05) between groups; they were lower in E- brains and contained DHA with DHA-PC 38:6 at the highest abundances. Moreover, hydroxy-DHA-PC 38:6 was increased in E- brains (P = 0.0341) supporting the hypothesis of DHA peroxidation. More striking was the depletion in E- brains of nearly 60% of 19 different lysophospholipids (lysoPLs) (combined P = 0.0003), which are critical for membrane PL remodeling. Thus, E- brains contained fewer DHA-PLs, more hydroxy-DHA-PCs, and fewer lysoPLs, suggesting that lipid peroxidation depletes membrane DHA-PC and homeostatic mechanisms to repair the damage resulting in lysoPL depletion.

methanol:water (80:20, v/v). Samples were extracted using a modifi ed Bligh and Dyer solvent system ( 16 ); the fi nal ratio was chloroform:methanol:water (2:2:1.8). Samples were mixed vigorously for 20 s, and then centrifuged at 1,000 g for 10 min, the top layer was removed by aspiration and discarded. The bottom layer was collected and evaporated under nitrogen, then stored at Ϫ 80°C less than 24 h before MS/MS ALL TM analysis. MS/MS ALL TM analysis was performed using the TripleTOF 5600 mass spectrometer with direct infusion into ESI source (AB SCIEX, Framingham, MA). Each sample was reconstituted with 400 l methanol:chloroform (1:2, v/v) containing 5 mM ammonium formate. Samples were delivered to the source at 15 l/ min using an isocratic fl ow of methanol:isopropanol (3:1, v/v) with 5 mM ammonium formate. A second isocratic pump delivered a solution of 98% isopropanol and 2% methanol containing 5 mM ammonium formate, as a make-up fl ow to the source through a T-junction at a rate of 60 l/min. Total fl ow was 75 l/min at point of entry into the ESI source. Source parameters included nebulizing gases: GS1 at 20 psi, GS2 at 15 psi, curtain gas at 20 psi, positive mode ion spray voltage at 5,500 V, negative mode ion spray voltage at Ϫ 4,000 V, declustering potential at 40 V, and at an ESI source operating temperature of 400°C. Collision energy for each MS/MS step was 50 ± 30 eV and Ϫ 40 ± 30 eV, respectively, for positive and negative ion mode experiments.
A fully automatic workfl ow was developed using the infusion MS/MS ALL TM technique on the TripleTOF 5600 system with a fl ow injection sample introduction strategy. The MS/MS ALL TM infusion method ( 17 ) allowed automated collection of TOF MS and TOF MS/MS of all lipids, using a step-wise sampling of precursor ions selected at unit resolution (1 amu) in quadrupole (Q1), followed by fragmentation in Q2, and then product ions were detected simultaneously using TOF. High-resolution TOF MS/MS spectra were collected with an accumulation time of 250 ms from lipid precursors from mass range m/z 200.050 to 1,200.049, followed by 1,000 product ion experiments with 1,000 precursors evenly spaced from m/z 200.050 to m/z 1,200.049. The data were accumulated for 100 ms for each fragment and collected in order from low to high m/z . The total time to carry out the MS/MS ALL TM acquisition for one sample was 5.48 min. Each sample was run in positive and negative modes. To obviate carryover, there was a washing time (5 min) between samples. Sample carryover was assessed during the wash step by injecting sample blanks containing only solvent.

Brain tissue extraction for DHA-PC analysis by triple quadrupole LC-MS/MS
The brains from E Ϫ and E+ zebrafi sh (n = 10 per group) were dissected and weighed. Brains were transferred individually to Eppendorf tubes, homogenized under 4°C, and extracted individually by adding, in the following order: 250 l methanol:water (1:1, v/v), DTD-PC as an internal standard (2.5 g in 5 l methanol), and 750 l methylene chloride:isopropanol:methanol (25:10:65, v/v/v). Solvents were all at 4°C. Samples were mixed vigorously for 20 s, and then centrifuged at 13,000 g for 10 min. The extracts were transferred to a new tube and were stored less than 24 h at Ϫ 80°C until analysis.
LC was carried out using a Shimadzu high-pressure LC system (Columbia, MD) with LC elution conditions as described previously ( 18 ). The system was coupled to an Applied Biosystems API 3000 triple quadrupole mass spectrometer with a TurboIon spray source operated with both mass analyzers set at unit resolution in positive mode (LC-MS/MS, Applied Biosystems/MDS Sciex API 3000, Foster City, CA). Nebulizer, curtain, and collision (CAD) gas parameters were set at 8, 10, and 2 psi, respectively. Heater gas was supplied at 6 l/min at 425°C. All gases were high purity Purkinje cells in brains from ␣ -TTP knockout mice. The specifi c lipids depleted during vitamin E defi ciency, however, are not known.
The objective of this study is to elucidate the molecular consequences of vitamin E defi ciency on lipids in zebrafi sh brains. However, embryonic brains are too small for rigorous analysis; therefore, we used adult brains from E Ϫ compared with E+ zebrafi sh to evaluate changes in lipid composition and abundance, and applied a combination of targeted and untargeted methods to enable a comprehensive profi ling of brain lipids to test this hypothesis. We chose zebrafi sh nearly 1 year of age because long-term vitamin E defi ciency causes severe depletion of tissue ␣ -tocopherol, resulting in behavioral and physiological damage ( 5,9 ). It should be noted that the diets fed were adequate in ascorbic acid, but restricted in long chain PUFAs. These 1-year-old zebrafi sh are equivalent to middle aged ( 10, 11 ), but not elderly, adult humans eating a low vitamin E diet ( 12 ).

Zebrafi sh housing and feeding
This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The Institutional Animal Care and Use Committee of Oregon State University approved the protocol (ACUP number 4344). Every effort was made to minimize suffering; prior to sampling, all fi sh were euthanized by an overdose of tricaine (MS 222; Argent Chemical Laboratories, Inc., Redmond, WA).
Tropical 5D strain zebrafi sh were housed in the Sinnhuber Aquatic Research Laboratory. Adult zebrafi sh were kept under standard laboratory conditions at 28.5°C with a 14 h light/10 h dark cycle. For 9 months, beginning at 50 days postfertilization, zebrafi sh were fed defi ned diets, which contained only fatty acids shorter than 18 carbons with two or three double bonds ( 2,5,9 ), were prepared in 300 g batches with the vitamin C source as StayC (500 mg/kg; Argent Chemical Laboratories Inc.) without (E Ϫ ) or with (E+) added vitamin E (500 mg RRR -␣ -tocopheryl acetate/kg diet; ADM, Decatur, IL), as described previously ( 5 ). Diets were stored at Ϫ 20°C until fed to the zebrafi sh.
Using high-pressure LC with electrochemical detection, ␣tocopherol was measured in diets, brains, and the matching bodies (without brain), as described ( 13 ); vitamin C was measured in diet as described ( 14 ). Measured ␣ -tocopherol concentrations in the E Ϫ and E+ diets were 1.6 ± 0.1 mg/kg and 334 ± 12 mg/kg, respectively; vitamin C was 143 ± 16 mg ascorbic acid/kg. This level of dietary vitamin C has been found to be adequate for the zebrafi sh ( 5,15 ).

Zebrafi sh tissue extraction and lipidomic analysis by MS/MS ALL TM
For survey experiments using MS/MS ALL TM , the head was removed just behind the eye from whole frozen zebrafi sh (n = 5 per group). The heads were weighed, then homogenized in P < 0.05. To discover signifi cant features with the greatest changes between the E Ϫ and E+ diet groups (n = 5 per group), the resulting P values (Student's t -tests) were plotted against fold change (log 10), calculated by MarkerView. Features with both the largest fold changes and smallest P values were selected for identifi cation.
For quantitative data from the LC-MS/MS experiments, statistical analyses of DHA-PC concentrations were performed using GraphPad Prism software (GraphPad, La Jolla, CA). Signifi cance ( P < 0.05) was determined using logarithmically transformed data by a one-way ANOVA followed by a Tukey post hoc test. To estimate overall signifi cance of lysophospholipids (lysoPLs), the responses for each individual were summed and the sums compared using a t -test.
The data from UPLC-TOF-MS/MS lipidomic analyses were imported into PeakView software for relative quantifi cation and lipid identifi cation. Lipid species were confi rmed by highresolution MS, MS/MS fragmentation, and isotopic distribution, and then compared using the PeakView database. Peak intensities were used for relative quantifi cation between E+ and E Ϫ zebrafi sh brains and were corrected for added internal standard (DTD-PC) and brain weight of each brain. Statistical differences between lipids were assessed as described for LC-MS/MS experiments.

Discovery of differentiating lipidomic metabolites and identifi cation by MS/MS ALL ™ analysis
We used an untargeted shotgun lipidomics method consisting of the ordered acquisition of high-resolution accurate mass TOF detection to identify several characteristic lipid species, including PC, lysophosphatidylcholine (LPC), phosphatidylethanolamine (PE), SM, phosphatidylglycerol (PG), and TGs. Principal component analysisdiscriminant analysis revealed signifi cant differences between the lipid profi les of the extracts of E Ϫ and E+ zebrafi sh heads. The score plot provides a visual distribution of the features and degree of discrimination between the E Ϫ and E+ extracts ( Fig. 1A ), while the loading plot shows a number of lipids that lead to sample clustering ( Fig. 1B ). PC 38:6 was selected for further analysis because it both decreased in E Ϫ extracts ( P < 0.05) and showed large fold differences between the two groups ( Fig. 1C ).

Brain DHA-PC (PC 16:0/22:6) and vitamin E concentrations
Brains were excised from individual fi sh; extracts were analyzed for PC 38:6, which was identifi ed as DHA-PC ( Fig. 2A, B ). Brain PC 38:6 identifi cation was based upon the retention time of an authentic standard and the detection of protonated molecular ion of m/z 806.7 [M+H] + (data not shown). Moreover, the collision-induced dissociation of molecular ion m/z 806.7 produced a characteristic fragment ion of m/z 184 corresponding to the PC moiety ( 19 ). DTD-PC, the internal standard, also showed the same fragment ion of m/z 184 ( Fig. 2A ). The fragment ion, m/z 790.5408 shows loss of a -CH 3 group from PC; the fragment ion, m/z 480.31882, is characteristic of 1-hexadecanoyl-sn -glycerophosphocholine ( Fig. 2B ). Additionally, nitrogen supplied by a custom liquid nitrogen system (Polar Cryogenics, Portland, OR). The ionizing voltage was 5,000 V, and the declustering, focusing, entrance, and exit potentials were 50, 200, 10, 35, and 12 V, respectively.
Multiple reaction monitoring (MRM) was used to quantify PC 38:6, which was identifi ed as DHA-PC in brain extracts . Analytes were detected using MRM, which allowed for simultaneous quantifi cation of specifi c isomers based upon unique transitions between the m/z of the precursor ion [M+H] + in the positive mode and characteristic product ions for each species, as well as their characteristic LC retention times. Authentic standards of DHA-PC and DTD-PC eluted at 14.6 min and 11.2 min, respectively. Measurements of DHA-PC limit of detection and limit of quantitation were assessed using the area ratio of DHA-PC to DTD-PC. The limit of detection and the limit of quantitation were defi ned as the concentrations with a signal-to-noise ratio of at least 3 and 10, respectively. The DHA-PC detection limit was 0.59 pmol/injection and the quantitation limit was 1.76 pmol/injection. The LC-MRM standard curves relating peak area ratios of DHA-PC/DTD-PC versus known authentic DHA-PC amounts (0.2-10 g injected) were used to quantitate DHA-PC; values are expressed per brain weight.

Ultra-performance LC-TOF-MS/MS lipidomic analyses
The brain tissue extraction procedure for untargeted lipidomics was the same as the extraction for DHA-PC, as described above. Brain weights were not signifi cantly different between groups (E Ϫ 2.3 ± 0.3 g compared with E+ 2.2 ± 0.4 g). To identify lipid metabolites, LC was performed using a 1.8 m particle 100 × 2.1 mm id HSS T3 column (Waters, Milford, MA) coupled to a quadrupole TOF mass spectrometer (AB SCIEX, TripleTOF 5600) operated in information-dependent MS/MS acquisition mode. The column was heated to 50°C in the column oven. A gradient system was used consisting of mobile phase A (60:40 acetonitrile:water containing 10 mM ammonium formate) and mobile phase B [90:10:5 (v/v/v) isopropanol:acetonitrile:water with 10 mM ammonium formate] with sample analysis performed over 17 min total run time. The initial starting conditions were 50% A and 50% B, then for 10 min the gradient was ramped in a linear fashion to 100% B and held at 100% B for 2 min. Subsequently, the system was switched to the initial ratio for 1 min, and equilibrated at the initial ratio for additional 4 min. The fl ow rate was 0.4 ml/min and the injection volume was 10 l. TOF MS acquisition time was 0.25 s, and MS/MS acquisition time was 0.1 s. The scan range was m / z 70-1,200 for TOF MS and m / z 50-1,200 for MS/MS. Source parameters included nebulizing gases: GS1 at 45, GS2 at 50, curtain gas at 35, positive mode ion spray voltage at 5,500 V, negative mode ion spray voltage at Ϫ 4,000 V, declustering potential at 80 V, and at an ESI source operating temperature of 500°C. Collision energy for each MS/MS step was 35 ± 10 eV and Ϫ 35 ± 10 eV, respectively, for positive and negative ion mode experiments. Data were generated from the ultra-performance LC (UPLC) separation of the lipid extract of individual brains, TOF accurate mass detection, and MS/MS fragment characterization.

Data processing and statistical analyses
The data acquired during the Infusion MS/MS ALL TM and LC-MS/MS experiments were processed in batch mode using LipidView software version 1.2 (AB SCIEX) for automated identifi cation and quantitation of lipid species. The processed Lipid-View data fi les were imported into MarkerView software (AB SCIEX) for initial data processing, including feature detection, peak integration, principal component analysis, and discriminant analysis.
Student's t -test comparisons were carried out between the two diet groups using MarkerView. Statistical signifi cance was set at compared with E+ bodies without brains (137 ± 13 nmol/g, <0.05, paired t -test).

Untargeted lipidomics identifi es alterations in brain lipids
To further examine alterations in lipid distribution and relative abundances between E+ and E Ϫ zebrafi sh brains, we used an UPLC lipidomics technique ( Fig. 3A, B ). Four lipid classes including 13 lipid species with 155 specifi c lipids were identifi ed ( Fig. 3C ) Quantitatively E Ϫ brains contained 30% less DHA-PC than did E+ brains ( P < 0.05, Fig. 2C ). Additionally, ␣tocopherol concentrations in E Ϫ brains (3.4 ± 0.1 nmol/g, mean ± SEM) were 70 times lower than in the E+ brains (243 ± 32 nmol/g); ␣ -tocopherol concentrations in E Ϫ bodies without brains (2.3 ± 0.3 nmol/g) were 60 times lower than in E+ bodies without brains (137 ± 13 nmol/g; two-way ANOVA: main effect of diet P < 0.0001, main effect of body part P = 0.0014). ␣ -Tocopherol concentrations in E Ϫ brains (3.4 ± 0.1 nmol/g) were higher than in the rest of the body (2.3 ± 0.3 nmol/g, P < 0.05); as were ␣tocopherol concentrations in E+ brains (234 ± 32 nmol/g) Fig. 1. Identifi cation of PC 38:6 as signifi cantly depleted in heads from E Ϫ zebrafi sh. For 9 months beginning at 50 days postfertilization, zebrafi sh were fed defi ned diets (E+ or E Ϫ ), which contained only fatty acids shorter than 18 carbons with two or three double bonds. For survey experiments using MS/MS ALL TM , the head was removed just behind the eye from whole frozen zebrafi sh (n = 5 per group). The heads were weighed, then homogenized in methanol:water (80:20, v/v). Samples were extracted using a modifi ed Bligh and Dyer solvent system ( 16 ); the fi nal ratio was chloroform:methanol:water (2:2:1.8). Samples were mixed vigorously for 20 sec, and then centrifuged at 1,000 g for 10 min, the top layer was removed by aspiration and discarded. The bottom layer was collected and evaporated under nitrogen, then stored at Ϫ 80°C less than 24 h before MS/MS ALL TM analysis. A: Score plot provides a visual distribution of the features and degree of discrimination between individuals in E Ϫ compared with E+ groups. B: Loading plot shows the lipids that lead to sample clustering. Positive percent loading shows higher and negative percent loading shows lower intensities of lipid metabolites in the E+ group. C: Volcano plot of the Student's t -test versus the log fold-change between groups. PC 38:6 (circled) was signifi cantly decreased in E Ϫ extracts. TAG, triacylglyceride. patterns. Depending upon collision energy, a head group and/or one or more fatty acid chains were lost from the lipid molecules, which allowed us to identify the lipid class and fatty acid chains. PE generated the characteristic losses of 43 and 140 Da known to be associated with PEhead group losses; PS showed Ϫ 87 and Ϫ 184 Da loss. The lysoPL abundances varied with regard to fatty acid, but were uniformly higher in E+ brains. lysoPLs that were signifi cantly higher in E+ and present at the highest abundances included LPC 16:0, LPC 22:6, and LPE 22:6 (top row, Fig. 6 ). There were several lysoPLs with saturated fatty acids signifi cantly higher in E+ compared with E Ϫ brains; these not only included LPC 16:0, but also LPC 18:0, LPE 18:0, and LPS 18:0 and at lower abundances LPE 16:0 and LPE 20:0. Overall, the lysoPLs decreased in E Ϫ brains by nearly 60% ( P = 0.0003).
To determine whether loss of PC 38:6 (16:0/22:6) in E Ϫ brains could be a result of lipid peroxidation, the relative abundances of oxidized-DHA-PC in the extracts were assessed ( Fig. 5  Remarkably, 19 lysoPLs were present at signifi cantly higher concentrations in the E+ compared with E Ϫ brains ( Fig. 6 ). These lysoPLs have been identifi ed and confi rmed using exact mass matching and MS/MS fragment located at the sn -1-acyl moiety and PUFAs are located at the sn -2-acyl moiety; the latter are released by phospholipase A (PLA) 2 ( 27 ). Hypothetically, an oxidized DHA would be released by this mechanism, generating a lysoPL with a saturated or monounsaturated fatty acid. This mechanism would explain a higher level of lysoPLs with saturated fatty acids in E Ϫ brains; however, we observed the opposite, lower levels in E Ϫ brains of all the lysoPLs that were different between the two groups. An alternative explanation is the involvement of PLA 1 enzymes releasing the sn -1-acyl moiety of PLs ( 28 ). Kuge et al. ( 29 ) investigated the role of PLA 1 in regulating the localized distribution of PL species in the plasma membrane of cultured neurons and suggest that PLA 1 may also contribute to the dynamic remodeling of the neural membrane.
Although our lipidomics analyses do not allow determination of the position of the fatty acids, we found three DHAcontaining lysoPLs (LPC, LPE, and LPS) that were present at signifi cantly lower concentrations in the E Ϫ brains. In P = 0.0003). The wide variety of lysoPLs that is depleted suggests that the entire lysoPL population is affected. lysoPLs are needed for PL remodeling during membrane synthesis, repair, and replacement ( 20,21 ). Because rates of DHA synthesis in the brain are insuffi cient to meet brain DHA requirements ( 6,22,23 ), it has been proposed that the brain acquires DHA as docosahexaenoyl-sn -glycero-3-phosphocholine (lyso-DHA-PC) ( 24 ). Recently, a transporter from the major facilitator superfamily, major facilitator superfamily d2a (MFSD2a), which has high specifi city toward lyso-DHA-PC ( 25 ), was shown to function as a DHA delivery mechanism facilitating brain docosahexaenoyl-sn -glycero-3-phospholipid (lyso-DHA-PL) uptake ( 25 ). Additionally, the MFSD2a transporter is critical to maintain the blood-brain barrier ( 26 ).
These major effects of vitamin E status on brain lysoPLs are surprising because lysoPLs with both saturated and unsaturated fatty acids are depleted in the E Ϫ brains. Generally, PLs have saturated fatty or monounsaturated acids Fig. 4. Four specifi c PLs containing DHA are signifi cantly lower in E Ϫ brains. Data were generated from the UPLC separation of the lipid extracts of individual brains from E Ϫ and E+ fi sh, extracted as described in Fig. 2 . Lipid species were confi rmed by high-resolution MS, MS/MS fragmentation, and isotopic distribution, and then compared using the PeakView database; PeakView software was also used for relative quantifi cation and lipid identifi cation. Peak intensities were used for relative quantifi cation between E+ and E Ϫ zebrafi sh brains. Shown are the individual values (E Ϫ , circles; E+, squares; n = 10 per group) with the median indicated with a line. was in the 1 or the 2 position because we lack authentic compounds; however, the high accuracy mass spectrometry data supports the identifi cation of hydroxy-DHA-PC. Based on the relative responses of DHA-PC (16:0/22:6, Fig. 4 ) and hydroxy-DHA-PC ( Fig. 5C ), hydroxy-DHA-PC is less than 0.2% of the DHA-PC.
Our fi ndings suggest that increased lipid peroxidation due to inadequate ␣ -tocopherol leads to the depletion of a highly abundant brain PL, DHA-PC, as well as three other PLs containing both DHA and oleic acid (18:1) (PC, PE and PS, Fig. 4 ). Given ␣ -tocopherol's role as a peroxyl radical scavenger ( 31 ), it is expected that PLs containing the highly unsaturated DHA would be depleted from the E Ϫ brains. dogs, supplements of fi sh oil (0.45, 0.9, or 1.35% fat) or vitamin E ( dl -␣ -T-acetate, 640 IU/kg diet) increased serum 1-docosahexaenoyl-sn -glycero-3-phosphocholine concentrations ( 30 ). This fi nding of DHA specifi cally in the sn-1 position further supports our fi ndings of decreases in LPC-DHA in E Ϫ adult zebrafi sh brains and the role of lyso-DHA-PC as the form DHA is transported from the liver to the brain ( 25 ).
DHA-PC was the lipid that showed the largest folddifferences between brains from E+ and E Ϫ zebrafi sh. Independent quantitation of DHA-PC showed that it was 30% lower in E Ϫ brains. Moreover, hydroxy-DHA-PC was elevated ( Fig. 5 ). We have not identifi ed whether the hydroxy-DHA  ). Previously, we identifi ed a more rapid depletion of both DHA and AA from E Ϫ zebrafi sh embryos ( 5 ), but in these adult brains, we found little evidence for AA depletion. This difference may be a result of the greater likelihood for oxidation of DHA ( 32 ). Alternatively, the brain may be able to synthesize suffi cient AA, but not DHA, for its needs ( 33 ). Additionally, 7-dehydrocholesterol, an intermediate in cholesterol synthesis, is more highly oxidizable than either of these fatty acids, but using our lipidomics data, we did not detect either 7-dehydrocholesterol or its oxidation product. It is likely that adult brain does not synthesize cholesterol extensively and therefore 7-dehydrocholesterol was not detectable.
Neural membrane PL remodeling liberates both AA and DHA, which serve as precursors for lipid mediators and signaling molecules. For example, AA can be metabolized to infl ammatory eicosanoids via cyclooxygenases to yield prostaglandins and thromboxanes or via lipoxygenases to yield leukotrienes ( 34 ). By contrast, DHA can be metabolized to the anti-infl ammatory docosanoids via lipoxygenase enzymes to yield D-series resolvins and other neuroprotective compounds (e.g., neuroprotectin D1) ( 35 ). Given the high brain DHA concentrations, as well as DHA's role in supporting neurological health across the lifespan ( 36,37 ), it is remarkable that PL-containing DHA, but not AA, was different between adult E Ϫ and E+ zebrafi sh brains. These results suggest that ␣ -tocopherol is required within the nervous system to protect DHA and thereby maintain the brain's homeostatic balance between anti-infl ammatory and infl ammatory compounds. These results also show that there was not artifactual nonspecifi c lipid peroxidation of the E Ϫ samples.
We are pursuing vitamin E defi ciency effects in development of zebrafi sh embryos. Notably, Huang et al. ( 38 ) have published that PC 38:6 is decreased in 24 and 48 h embryos compared with earlier time points. These fi ndings suggest that PC 38:6 is an important target to evaluate in the E Ϫ embryo.
We report herein that ␣ -tocopherol defi ciency depletes four DHA-containing PLs and 19 lysoPLs. A unique biological role has been proposed for vitamin E in membrane repair ( 39 ). Potentially, the alterations in lysoPLs we observed in the E Ϫ brains are a result of a depletion of DHAcontaining PLs and increased requirements for DHA for PL synthesis and lysoPLs for remodeling for membrane repair ( Fig. 7 ). Presumably, inadequate ␣ -tocopherol concentrations allow lipid peroxidation to deplete not only brain DHA-PC, but DHA throughout the body, thereby limiting DHA delivery to the brain. Our data delineate that critical lipids are protected by ␣ -tocopherol and suggest why ␣ -tocopherol is needed as a vitamin.
The authors thank Carrie L. Barton and Jane La Du for providing excellent technical assistance. , which is reduced to DHA-(OH)-PC (hydroxy-DHA-PC was detected in this study). ␣ -Tocopherol ( ␣ -T) limits this process. Inadequate vitamin E allows lipid peroxidation to damage the membrane, which must be repaired. In the face of inadequate ␣ -T, more lysoPLs are needed for membrane repair, but sufficient amounts, especially of lyso-DHA-PC, may not be available (possibly due to increased lipid peroxidation in the rest of the body). The increased membrane remodeling depletes various lysoPLs in the brains of E Ϫ zebrafi sh.